Characteristics of phantom limb pain in U.S. civilians and service members

Study design

The study design was cross-sectional and observational. Data came from a study that collected survey data and blood samples from volunteers to identify genetic variations associated with PLP. The study was approved by the Walter Reed National Military Medical Center Institutional Review Board in Bethesda, MD, IRB00008418. The survey asked about demographics, cause of limb loss, experience and characteristics of PLS and PLP, potentially related factors such as pre-amputation pain, and other health concerns.

Setting

Participant recruitment for the present sample occurred from 2012 to 2016 at several sites, including United States Military Health Facilities (MHF), university health centers, and events hosted by the Amputee Coalition of America. Recruitment was conducted by placing flyers or banners at the facilities or by approaching potentially eligible individuals being treated at medical facilities.

Participants

Adult participants were considered eligible if they had at least one upper or lower limb amputation for at least three months prior to enrollment and could follow study instructions. Participants were excluded if they had known uncontrolled systemic disease (not including diabetes mellitus), if they had any condition (or medication) that would confound study results or make a blood draw inadvisable, or if they were unable to fill out the phantom limb pain questionnaire. Initially the study only included individuals of European-American descent to reduce heterogeneity for the genetics study, but this criterion was removed on 01/31/2014. All participants provided written informed consent.

Measures

Participants self-reported their responses on a paper or electronic questionnaire, and research assistants were permitted to explain questions if participants did not understand them. The PLP questionnaire collected data on demographics, limb loss location and cause, temporal characteristics of PLS and PLP, magnitude of recent PLP (i.e., level of PLP in the past 24 h), average pain (i.e., level of pain for an average episode of PLP), and worst ever pain (i.e., worse level of PLP ever experienced), presence of telescoping, and control of phantom limb movement. Participants quantified their level of pain on a visual analog scale (range: 0–100), interpreted respectively as “No pain” and “So much pain I would faint.”

Analyses

Descriptive analyses included the reporting of onset, frequency, and duration of PLS and PLP episodes. Inferential analyses included generalized linear regression and analysis of variance (ANOVA) to assess the relationship between characteristics of PLP and time since amputation, and certain factors such as having experienced telescoping. Planned contrasts were conducted to examine whether those with congenitally missing limbs reported less PLP (recent, average, and worst ever) than the other groups combined. Analyses were conducted using Statistical Package for the Social Sciences (SPSS; Version 23). SPSS defaults were used for missing data: complete case analysis. Continuous variables were assessed for skewness and kurtosis using the SPSS outputted values (acceptable range ± 2).

Participants

From March 2012 to November 2016, there were 699 screenings for this study. Thirteen of those screened were ineligible. Of the 686 eligible screenings, 74 turned out to be duplicates; i.e., some participants enrolled twice in the study. Only the first data entry for each duplicate subject (n=37) was used for data analyses, generating a total analyzable sample of 649. See Table 1 for a summary of sample characteristics. The sample was predominately male (65%) and white (83.3%). The mean age was 47 years (range: 18–83). More information on the sample (to include limb loss location and cause) is included in Table 2.

Prevalence and onset

Eighty seven percent (n=566) of the sample (n=649) reported having experienced PLS; 82% (n=531, or 94% of those with PLS) reported having experienced PLP. See Table 3 for a summary of the onsets of PLS and PLP reported by participants. The most common onset for both PLS and PLP was immediately after amputation (47% of participants).

Chronicity

To examine the chronicity of PLP, respondents (n=611, includes participants who denied experiencing PLP) were stratified into the following groups according to time since their amputation: three months to a year (n=93), 1–2 years (n=104), 2–5 years (n=125), 5–10 years (n=116), and over 10 years (n=127). Participants with congenitally missing limbs (n=28) were excluded from these analyses.

Linear regressions were conducted to assess time since amputation (in months) as a predictor of recent PLP, average PLP, and worst PLP experienced. Time since amputation in months predicted recent PLP [R²=0.024, F(1,564)=13.77, p<0.001, β=−0.154, 95% CI [−0.23, −0.07], p<0.001], but not average PLP (β=0.028, 95% CI [−0.05, 0.11], p=0.513) or worst ever PLP (β=0.002, 95% CI [−0.08, 0.08], p=0.969).

Adjusting for age and gender did not change results. Time since amputation predicted recent PLP [R²=0.031, F(3,562)=6.03, p<0.001, β=−0.168, 95% CI [−0.05, −0.02], p<0.001] after accounting for age (β=0.038, 95% CI [−0.08, 0.21], p=0.389) and gender (β=0.069, 95% CI [–0.82, 8.41], p=0.106).

Frequency, duration, and intensity of PLP episodes

The frequency and typical duration of PLP episodes are summarized in Table 3. On a scale of 0–100, the mean PLP over the past day was 17.25 (standard deviation [SD] 25.73), the mean average PLP was 36.84 (SD 28.00), and the mean worst PLP ever experienced was 62.14 (SD 35.98).

Telescoping

Twenty-eight percent (140 out of 508) of participants who reported PLS experienced telescoping (see Table 3). Of those who experienced telescoping 15% (21 out of 139) reported being able to control this telescoping.

Simple linear regressions were run to examine telescoping as a predictor of recent PLP, average PLP, and worst PLP ever experienced. Those who experienced telescoping reported higher levels of PLP over the past day [R²=0.025, F(1, 506)=12.98, β=0.158, 95% CI [0.07, 0.25], p<0.001], higher levels of average PLP [R²=0.015, F(1, 505)=7.79, β=0.123, 95% CI [0.04, 0.21], p=0.005], and higher levels of worst PLP ever experienced [R²=0.036, F(1, 506)=19.14, β=0.191, 95% CI [0.10, 0.27], p<0.001].

Controlling for age and gender did not change results. The presence of telescoping was associated with higher recent PLP [R²=0.036, F(3, 504)=6.19, β=0.153, 95% CI [0.07, 0.24], p<0.001], higher levels of average PLP [R²=0.019, F(3, 503)=3.179, β=0.128, 95% CI [0.04, 0.22], p=0.004], and higher levels of worst PLP ever experienced [R²=0.044, F(3, 504)=7.702, β=0.187, 95% CI [0.10, 0.27], p<0.001] when controlling for age and gender.

Phantom limb movement

About 42.4% (n=227) reported never being able to move their phantom limb at will, 39.4% (n=211) reported always having been able to move their phantom at will, 12.9% (n=69) reported that they can sometimes move their phantom limb, 2.2% (n=12) reported that they were initially able to move the phantom but lost this ability over time, 3.2% (n=17) reported that they were initially unable to move their phantom but gained this ability over time. Of the participants who could not always move their phantom limb(s) (n=212), 47.2% (n=100) reported that this led to the sensation of paralysis, cramping, or pain.

Pre-amputation pain

Forty-four percent (283 out of 647) of the sample reported experiencing pain in the limb prior to amputation. Simple linear regressions were conducted to examine pre-amputation pain as a predictor of recent PLP, average PLP, and worst PLP ever experienced. Higher levels of pre-amputation pain were associated with higher levels of recent PLP [R²=0.011, F(1,636)=6.95, β=0.104, 95% CI [0.03, 0.18], p=0.009] and higher levels of worst PLP experienced [R²=0.013, F(1,635)=8.60, β=0.116, 95% CI [0.05, 0.25], p=0.003]. However, there was not support for pre-amputation pain as a predictor of average levels of PLP (β=0.065, 95% CI [−0.01, 0.14], p=0.102).

Controlling for age and gender did not change results. Higher levels of pre-amputation pain were associated with higher levels of recent PLP [R²=0.014, F(3,634)=3.085, β=0.092, 95% CI [0.01, 0.17], p=0.027] and higher levels of worst PLP experienced [R²=0.016, F(3,633)=3.489, β=0.110, 95% CI [0.03, 0.19], p=0.007], but not average levels of PLP (β=0.054, 95% CI [−0.03, 0.13], p=0.189).

Cause of limb loss

Table 2 contains the number and percentage of participants according to reason for amputation/absence of limb(s) and the mean PLP (recent, average PLP, and worst PLP). One-way analyses of variance were run to look at cause of limb loss as a predictor of recent PLP, average PLP, and worst ever PLP. There was a main effect of reason for limb loss on average PLP [F(4,595)=8.361, p<0.001] and worst ever PLP [F(4,597)=12.83, p<0.001] but not recent PLP [F(4,598)=2.34, p=0.054].

Planned contrasts were conducted to examine whether those with congenitally missing limbs reported less PLP (recent, average, and worst ever) than the other groups combined. Those with congenitally missing limbs reported lower levels of recent PLP [t (37.93)=3.93, p<0.001], lower average PLP [t (37.22)=7.43, p<0.001], and lower worst ever PLP [t (30.73)=5.96, p<0.001]. PLS and PLP were nevertheless present in the sample of individuals with congenital limb loss, as 43% (n=12) reported having experienced PLS, and 36% (n=10) reported having experienced PLP.

Prevalence and onset

This study found that 87.2% of the sample reported having experienced PLS and 81.8% reported experiencing PLP. These rates fall on the higher end of the range reported by earlier research (PLS: 76–100% and PLP: 51–85%; [4], [5], [6]). These estimates highlight the pervasiveness of PLP, underlining the importance of identifying effective treatments for PLP as the population of persons with limb loss rises.

Participants reported nearly identical onset for both PLS and PLP, indicating that the two coincide. The onset of PLS/PLP varied widely across subjects but tended to occur earlier rather than later: roughly half of participants experienced phantom limb phenomena immediately after amputation and approximately 70% of participants experienced the phenomena within a week of amputation.

This variability in onset is consistent with the literature, which has found that PLP occurs within the first few days of amputation for the majority of patients [4, 17, 20] but can also occur decades after amputation [5]. This study bridges these previously disparate findings, underlying the heterogeneity inherent to PLP.

Chronicity

The more time that had elapsed since the amputation the lower the recent PLP severity. However, time since limb loss did not predict the average PLP or the worst PLP ever experienced after excluding participants with congenitally missing limbs, who are likely to be further out from date of limb loss and may be less likely to experience PLP [21], [22], [23], [24]. Additionally, many participants continued to experience PLP years after their amputation: approximately 53% of participants who were 10 years out from their amputation reported having experienced PLP in the last month.

These findings indicate that over time the chance of experiencing PLP on given day decreased, but the majority of patients continue to experience PLP for years after an amputation. Moreover, there is no evidence that the intensity of an average episode of PLP decreases over time; that is to say: PLP becomes less frequent but not less intense. These findings bridge past studies showing that PLP fades yet endures: generally PLP decreases over time [7, 8] but lingers for the majority of patients [6, 9, 25, 26].

Frequency and duration of PLP episodes

This variability in both the frequency and duration of PLP episodes (Table 3) is consistent with research by Richardson, Glenn, Nurmikko, and Horgan [27]. This study builds upon this research by providing estimates using a more diverse sample of persons with limb loss: Richardson et al.’s sample was limited to participants with peripheral vascular disease at six months post-amputation.

Telescoping

Of participants who experienced PLS, 28% reported having experienced telescoping, the sensation of the phantom limb retracting into the residual limb, which is slightly higher than the prevalence rates found in other studies (for a review see [28]). Only a very small percentage (3.2%) of these participants reported control over the telescoping—i.e., the ability to expand or retract the limb at will. Telescoping was associated with higher levels of PLP. This finding is consistent with more recent research by Grüsser et al. [13]and Montoya et al.[29], but contradicts earlier work [10], [11], [12] asserting that telescoping protected against PLP (see Montoya et al. [29] for a critique of the research methods used in this earlier work). Telescoping appears to be a risk, rather than a protective, factor for PLP.

Phantom limb movement

The majority of participants reported either always (42.4%) or never (39.4%) being able to move their phantom limb at will. About 12.9% of participants reported that the ability to move their phantom limb comes and goes, whereas a minority of participants reported losing (2.2%) or gaining (3.2%) the ability to move the limb over time. Of the participants who could not always move their phantom limb(s), 47.2% reported that this led to the sensation of paralysis, cramping, or pain.

The ability to move a phantom limb ties into the theory of learned paralysis [2, 30]. According to this theory, there are two means by which a phantom limb can become paralyzed. In the first, the brain learns that the limb is paralyzed before amputation; the motor cortex transmits signals to the move the limb but then receives visual feedback that the limb has not moved, leading to the sensation of paralysis. After amputation the brain continues to perceive the limb as paralyzed. In the second, the brain learns that the limb is paralyzed after amputation; the motor cortex transmits signals to move the limb but then does not receive visual feedback that the limb has moved, leading to the sensation of a paralyzed phantom limb. This study provides partial support for the theory of learned paralysis: approximately half of participants who were unable to move their phantom limb reported that this in turn led to pain. However, the theory does not account for the remaining participants who experience PLP. Moreover, only 2% of participants reported that their ability to move the phantom disappeared over time; the majority (82%) either always or never had the ability to move their phantom. These findings contradict the assertion that many if not all patients can move their phantom limb immediately after a non-traumatic surgical amputation but loses this ability over time [2].

Pre-amputation pain

The level of pre-amputation pain predicted the level of recent PLP and the worst level of PLP ever experienced, but not the level of PLP for an average episode. Overall, these findings indicate that higher levels of pain in the limb before amputation are associated with higher levels of subsequent PLP. These findings are consistent with several studies reporting a positive correlation between the presence of pre-operative pain in the limb and subsequent development of PLP [8, 15], [16], [17], [18], and collectively these findings indicate that pre-amputation pain is a risk factor for PLP. It is likely that past studies that did not detect an effect [6, 11, 31] were limited by statistical power.

Cause of limb loss

Those with congenitally missing limbs reported markedly lower levels of PLP (in terms of recent PLP, average PLP, and worst PLP) than those who had undergone an amputation. Otherwise, there were no consistent differences in PLP according to cause of limb loss.

The finding that participants with congenitally missing limbs reported less PLP is consistent with past research, which indicates that those with congenital limb loss experience less severe PLP [21], [22], [23], [24]. However, unlike in these studies—yet consistent with other reports in the literature documenting PLP in those with congenital limb deficiency [30, 32], [33], [34], [35]—phantom limb phenomena were still commonly reported even by participants with congenitally missing limbs: 45% of this group reported having experienced PLS and 36% reported having experienced PLP. It is likely that the experience of PLP in congenitally missing limbs has been underestimated and therefore undertreated.

The otherwise lack of consistent differences in PLP according to cause of limb loss is in accord with Sherman & Sherman [6], which did not detect differences in PLP according to reason for amputation in a sample of 764 participants. However, unlike the Sherman & Sherman study, in which 91% of participants had sustained traumatic amputations, the present study had a higher percentage of participants with non-traumatic amputations, and thus was better powered to speak to potential differences in PLP and is generalizable to the present population of Americans with amputation. Together, these studies indicate that reason for amputation is not a meaningful predictor of subsequent PLP.

Strengths and limitations

It is important to interpret findings in light of this study’s strengths and limitations. This study surveyed a large sample of persons with missing limb(s) with a range of etiologies, strengthening its ability to speak to the broader experience of persons with missing limbs. Furthermore, this study assessed a wide range of phantom limb phenomena, to include the topic of phantom limb movement which has been understudied to date despite its importance to PLP treatment.

The study also has several limitations. First, the study used self-reported survey data, relying upon each participant’s recollection of his or her health history and pain experience. There is evidence supporting the reliability of retrospective pain reports [36]. However, the possibility of inaccuracies remains, in particular for details of the initial PLP experience for participants who had undergone amputation(s) several years prior. Second, many of the measures for this study consisted of single-item questions rather than validated scales. Third, some questions were omitted or added later in the study, leading to smaller sample sizes for some analyses. Finally, the study used a convenience sample which may differ from the population of Americans with limb loss. In particular, the population of persons with amputation(s) due to diabetes may be underrepresented in this sample. However, this study had one of the most representative samples in the PLP literature due to the fact that it recruited participants across the United States who had experienced limb loss due to a wide range of causes.